Randomized Trial of Interventions to Improve Warfarin Adherence: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Randomized Trial of Interventions to Improve Warfarin Adherence (NCT00904982), a randomized na study. The primary outcome is Improved Warfarin Adherence/% Timeout of Target INR Range.

2. Clinical Question

3. Trial Design

This was a completed, randomized, parallel-group trial sponsored by the University of Pennsylvania. The registry lists an enrollment of 268, with a study start in 2009-10 and primary completion in 2012-05. The phase is recorded as NA. The study title is Randomized Trial of Interventions to Improve Warfarin Adherence.

Participants were allocated to four groups: usual care/control, Med-eMonitor with active reminders, lottery incentive, or the combined lottery-and-reminder intervention. The brief description says group selection was randomized by the study computer. The registry classifies masking as single, but the identity of the masked party is not reported in the registry record. The allocation concealment process is also not reported in the registry record.

The Med-eMonitor was described as an electronic medication monitoring system for home use. In the reminder group, it was set to active mode to display encouraging messages. The lottery group was entered into a daily lottery. The combined group received both features. The control description indicates use of the monitor without the active reminder mode or lottery. More detailed descriptions of how the usual-care background differed across groups are not reported in the registry record.

The primary endpoint was assessed over six months. The registry notes that the primary analysis required multiple INR results and excluded participants for whom those results were unavailable. The number of participants included in each primary analysis is posted by group, but the registry record does not give a fuller account of participant flow, withdrawals, adherence to assigned interventions, or the timing of outcome assessments.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Overall Study

Reasons for not completing this period:

Analysis population for the primary outcome, as defined in the registry: This analysis required multiple INR results. If those were not available for a subject, they were excluded from the primary analysis.

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as Improved Warfarin Adherence/% Timeout of Target INR Range, measured over six months.

The primary endpoint is titled Improved Warfarin Adherence/% Timeout of Target INR Range. It measures the percentage of time that a participant’s INR was outside the target range during the six-month assessment period. INR is a measure used to monitor anticoagulation. Time outside the target range describes the extent to which measured anticoagulation values were not within the intended range, rather than directly measuring medication-taking behavior.

The registry reports the endpoint as a median with an interquartile range for each group. A median describes the middle of the observed distribution, while the interquartile range describes the central spread of observations. These summaries can be useful when individual values are unevenly distributed or when extreme values might make a mean less representative. For this endpoint, a lower percentage would describe less time outside the target range in the observed group; that descriptive meaning alone does not establish a causal or clinically important difference.

Participants without multiple INR results were excluded from the primary analysis, according to the registry description. The number of such exclusions and the reasons for missing INR results are not reported in the registry record. The posted arm summaries therefore describe the participants with the required outcome data, rather than necessarily every randomized participant.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

Methods determine how a trial question is translated into an estimate and how much uncertainty should accompany that estimate. For this trial, the registry record describes randomization, a parallel-group design, a single-masked classification, and a median primary endpoint. It does not report a detailed statistical analysis plan, so the explanations below distinguish general statistical principles from the methods documented for this study.

What does randomization contribute?

In general, randomization assigns participants to study groups by chance. When implemented appropriately, it aims to make groups comparable at the start with respect to both measured and unmeasured characteristics. This supports causal comparisons, but randomization does not guarantee that every group will have identical baseline characteristics. It also does not correct problems that arise after assignment, such as missing outcome measurements or departures from the planned analysis.

The registry says that the study computer randomized participants among the four groups. The sequence-generation method, allocation concealment, and checks of baseline comparability are not reported in the registry record. The record classifies the study as single masked but does not say who was masked. Those details matter when considering possible influences on intervention delivery, outcome collection, or analysis.

Why use a median and an interquartile range?

In general, the median is the middle value after observations are ordered. It can represent the center of a distribution without being pulled as strongly by unusually high or low observations as an arithmetic mean can be. The interquartile range spans the central half of a distribution and gives a sense of how dispersed the observations are around the median. It does not describe the full range of individual values, nor does it by itself quantify uncertainty in the estimated median.

The registry identifies the primary parameter as a median and the dispersion as an interquartile range. The posted results report a median and two endpoints for the interquartile range in each group. No standard error, confidence interval for a group median, or confidence interval for a contrast is reported in the registry record. The spread among individual observations and the precision of an estimated group summary are different concepts.

How should group comparisons be assessed?

In general, a comparison between randomized groups should estimate a contrast that matches the clinical question and should report its uncertainty. With a median endpoint, possible methods may target differences in distributions or another clearly defined feature; the choice depends on the analysis plan and assumptions. A test result alone would not show the size or precision of an effect, so an estimate and interval are useful alongside any significance test.

The registry reports group-level medians and interquartile ranges but does not report the method used for inferential comparisons. It records zero analyses for this outcome. No p-value, confidence interval for a between-group contrast, adjusted estimate, or specified contrast is reported in the registry record. The reader therefore cannot determine from the posted results whether a formal analysis accounted for the four-arm structure or any covariates.

Why does missing outcome data matter?

In general, an analysis that excludes participants without outcome measurements can be biased if the availability of measurements is related to the outcome or differs according to group. The direction and size of any bias depend on why data are missing and on the values that would have been observed. Excluding incomplete cases can also reduce precision. Statistical methods for missing data rely on assumptions, and those assumptions should be explained rather than treated as known details.

For this trial, the registry says that multiple INR results were required and participants without them were excluded from the primary analysis. The number excluded, reasons for missing results, and any sensitivity analyses are not reported in the registry record. It is therefore not possible to assess the likely influence of missing INR data from the registry summaries alone.

What do planning calculations mean?

In general, sample-size planning uses assumptions about the effect to detect, the variability of the outcome, the significance level, allocation, and a desired power. A planning illustration based on an observed effect is not the trial’s observed power and does not replace a prospective design justification. The registry’s planning assumptions specify a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation, with the effect scaled to a stated fraction of the observed effect. The registry record does not report the trial’s own sample-size justification, and no planning sample-size estimates should be interpreted as observed results.

The registry also distinguishes planning calculations for time-to-event outcomes from other outcomes. The primary endpoint in this trial is a percentage summarized by medians, not a time-to-event endpoint. Without a reported inferential analysis for the primary outcome, the group summaries remain descriptive and cannot establish the precision or statistical significance of a treatment contrast.

10. Multiplicity, Missing Data and Analysis Populations

The trial has four randomized groups and a primary endpoint, creating several possible pairwise or combined comparisons. When multiple comparisons are examined, the chance of obtaining at least one apparently notable result by chance can increase. Statistical approaches can address multiplicity by prespecifying a limited set of comparisons or adjusting inference, but the appropriate approach depends on the scientific questions and analysis plan. The registry record does not report a multiplicity strategy or identify a primary contrast among the four groups.

The registry lists one outcome measure and zero analyses for that outcome. It also records zero secondary outcomes posted. As a result, the registry record does not provide inferential comparisons, subgroup analyses, or secondary outcome findings for this trial. This absence should not be read as evidence that no analyses were conducted; it means they are not reported in the registry record.

The primary analysis required multiple INR results, and participants without them were excluded. The record does not report missing-data counts by group, reasons for missingness, or sensitivity analyses. It also does not give a complete analysis-population description, such as an explicit intention-to-treat analysis or a per-protocol analysis. The posted outcome counts are therefore best understood as group-specific counts for the primary analysis, not as a complete account of all randomized participants.

In general, intention-to-treat analyses preserve comparison by randomized assignment, while other analysis populations can address different questions but may be more vulnerable to selection. Which population was used, and how missing observations were handled beyond exclusion for unavailable INR results, are not reported in the registry record. These omissions limit assessment of both the assumptions and robustness of the primary outcome summaries.

11. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 0% reporting threshold.

Most frequent serious adverse event terms

Most frequent non-serious adverse event terms

Serious events with the highest arm-level rate: major bleeding (7.6%), non-cns thrombosis (3.1%), stroke/tia (1.5%).

The registry reports serious adverse event counts by group. Serious events affected 4 participants in control, 4 in the reminder group, 4 in the lottery group, and 5 in the combined group. The corresponding numbers at risk were 65, 63, 65, and 66. These counts are not a statistical comparison and do not indicate whether observed differences arose by chance. The registry does not report deaths for these groups; deaths are not reported in the registry record.

For other adverse events, the counts were 1 in control, 3 in the reminder group, 2 in the lottery group, and 5 in the combined group. The corresponding numbers at risk were 65, 63, 65, and 66. The registry does not provide an inferential analysis of these figures or a fuller account of event collection and attribution. The counts should not be treated as rates, and no rates or between-group differences are calculated.

The listed leading serious adverse-event terms and percentages are Major Bleeding at 7.6%, Non-CNS Thrombosis at 3.1%, and Stroke/TIA at 1.5%. The registry record does not clarify the denominator or explain how these terms relate to the group-specific counts. These summaries describe reported safety information; they do not establish that an intervention caused or prevented an event.

12. What the Result Does, and Does Not, Mean

The primary endpoint addresses time outside the target INR range, an outcome related to anticoagulation control. The trial tested approaches intended to support medication adherence: electronic monitoring with active reminders, a daily lottery incentive, and their combination, alongside a usual-care/control group. The reported medians describe the observed outcome distribution among participants with the required INR data.

The posted results show that the median was 31.6 in control, 23.8 with reminders, 30.1 with the lottery, and 23.9 in the combined group. They also provide interquartile ranges for each group. These are useful descriptive summaries, but they do not give a formal estimate of the difference between groups or its precision. Because the registry record does not report a statistical comparison, it is not possible to use these numbers alone to determine whether the observed differences are distinguishable from chance variation.

The results do not establish that any strategy caused improved adherence or better anticoagulation control. The endpoint records time outside the INR target range, not medication-taking behavior directly. Nor do these summaries establish that the strategies had no effect. A lack of a reported inferential analysis is not evidence of equivalence, and similar-looking medians do not establish that groups were the same.

The interpretation is also limited by the primary analysis’s requirement for multiple INR results. Participants without those results were excluded, while the extent and pattern of missing data are not reported in the registry record. The identity of the masked party, the detailed analysis plan, and any sensitivity analyses are also not reported. These gaps prevent a full assessment of potential bias and statistical uncertainty.

The most supportable reading is therefore descriptive: the registry reports medians and interquartile ranges for the primary outcome across four randomized groups, with outcome-analysis counts for each group. It does not provide the inferential results needed to quantify comparative effects or support a conclusion about statistical significance or clinical importance.

13. Why This Trial Matters Statistically

This trial is statistically interesting because it compares several behavior-support strategies within one randomized study. A four-group design can help distinguish the descriptive pattern associated with reminders, incentives, and their combination. However, the design also makes it important to state which comparisons are central and how multiple comparisons are handled. The registry record does not report those decisions or the inferential analyses.

The outcome is a percentage of time outside the target INR range, summarized using a median and interquartile range. This emphasizes the distribution of observed values rather than an average alone. It also raises the need to interpret the measure carefully: it describes anticoagulation control, while the interventions were intended to influence adherence. A measure related to the intended pathway is not identical to direct measurement of the behavior itself.

The missing-data rule is especially relevant. The primary analysis excluded participants who lacked multiple INR results. Because the number and reasons for these exclusions are not reported, readers cannot judge whether the analyzed groups remain representative of their randomized groups. In a randomized trial, missing post-randomization outcomes can weaken the comparability that randomization is designed to support.

Finally, group summaries are not substitutes for estimates of comparative effects. The posted medians and interquartile ranges allow readers to describe the observed data, but no contrast estimates or uncertainty intervals are reported. This distinction is central to statistical interpretation: descriptive differences can motivate questions, while formal analyses are needed to quantify comparative uncertainty. The registry record leaves that inferential step unreported.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The registry record offers limited detail for evaluating how the primary analysis was conducted. It does not report a statistical analysis plan, a specified primary contrast among the four groups, a test or model for the outcome, or adjusted estimates. It also provides no confidence interval for a between-group effect and no p-value. Consequently, the precision and statistical significance of comparative effects cannot be assessed from the posted results.

The record says participants without multiple INR results were excluded from the primary analysis, but does not report how many were excluded or why. There is no reported comparison of participants with and without the required measurements, and no sensitivity analysis addressing missing outcomes. The effect of this exclusion on the group summaries is therefore uncertain.

Further trial details are also absent. The identity of the masked party, allocation concealment, baseline characteristics, participant flow, intervention adherence, and the precise procedures for measuring and deriving the percentage of time outside the target range are not reported in the registry record. The record does not describe how the target INR range was determined for participants or how repeated measurements contributed to the endpoint.

Safety reporting is limited to counts at risk and affected participants by group, plus a list of leading serious-event terms and percentages. The denominator and relationship between the listed terms and group counts are not explained. Without comparative safety analyses or fuller event definitions, the figures should be interpreted as registry summaries rather than estimates of causal risk.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Randomized Trial of Interventions to Improve Warfarin Adherence — NCT00904982.
• Registry record last updated 2017-03-16; status last verified 2017-01.
• Results first posted 2017-03-16.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

18. Record Summary

NCT00904982 is a completed randomized, parallel-group trial sponsored by the University of Pennsylvania. The registry lists 268 enrolled participants and a single-masked design. The study compared usual care/control with Med-eMonitor reminders, a daily lottery incentive, and the combined intervention.

The primary outcome was the median percentage of time outside the target INR range over six months, reported with an interquartile range. The posted medians were 31.6 for control, 23.8 for reminders, 30.1 for the lottery, and 23.9 for the combined group. Participants without multiple INR results were excluded from the primary analysis.

The registry record does not report inferential analyses for the primary outcome, a detailed analysis plan, or sensitivity analyses for missing data. It also reports adverse-event counts by group without a comparative statistical analysis. The posted findings therefore support descriptive interpretation of the analyzed groups, not a conclusion about comparative efficacy or safety.